Physicist review: Haisch-Bernard (US 7,379,286, "Quantum Vacuum Energy Extraction")

Role: PhD physicist (Step 2). Source tags: [read] from project documents or Investigator files, [web] with URL, [memory] with confidence, [calc] my own arithmetic (Python, 2026-10-04). Interested parties: Haisch and Moddel (inventors, Jovion Corporation stockholders; Moddel declares the stockholding in the 2019 paper) describe their own scheme; their statements are their characterization. No source gap: all three project documents are readable.

0. Summary of verdicts

1. Background the reader needs (plain language)

Zero-point energy (ZPE). A quantum harmonic oscillator of frequency f cannot have zero energy; its lowest energy is hf/2 (h is Planck's constant). The electromagnetic field is a collection of such oscillators ("modes"), so the vacuum has an unavoidable background of fluctuating fields, the "zero-point field" (ZPF). The spectral energy density per unit volume and frequency is proportional to the cube of the frequency, rho(w) = hbar w^3 / (2 pi^2 c^3), where w is angular frequency and hbar = h/2pi. [memory, high]

Casimir force. Two parallel, neutral, conducting plates a distance d apart attract with pressure P = pi^2 hbar c / (240 d^4). [memory, high; Forward's Eq. 2 with h in place of hbar is the same thing, checked by me below.] At d = 100 nm: P = 13 Pa (about 1.3e-4 atmosphere). [calc] Measured to a few per cent in many experiments since the late 1990s. [memory, high]

Casimir-Polder (CP) shift. A single atom near a wall (or between walls) has its ground-state energy lowered a little by the modified vacuum fluctuations. It is the same physics as the Lamb shift (the tiny shift of hydrogen's energy levels due to the vacuum field), but adjusted for the surfaces. In ordinary quantum electrodynamics (QED) this is a position-dependent energy U(z), which gives a force on the atom. It was measured in 1993 by sending ground-state sodium atoms through a micron-sized parallel-plate cavity; QED prediction for magnitude and distance dependence was confirmed. [web: https://digitalcommons.odu.edu/physics_fac_pubs/7/ (Sukenik, Boshier, Cho, Sandoghdar, Hinds, PRL 70, 560, 1993)]

Stochastic electrodynamics (SED). A research program (Marshall, Boyer, from the 1960s) that asks how much of quantum behaviour can be reproduced with classical particles plus a classical random ZPF. [read: Patent7379286 col.1-2]

2. Claim-by-claim review

Claim 1. "Casimir-Lamb shift": 1-10 eV lost per transit of a 0.1 micron cavity

The inventors' argument [read: Patent7379286 p.10-12; Haisch_abstractSSE10 p.1]. In the SED picture the hydrogen electron orbits at the Bohr radius a0 = 0.529 Angstrom, and its Larmor radiation (the energy any accelerated charge radiates) is balanced by absorption from the ZPF. The patent computes the wavelength of the relevant ZPF as "220.529137 = 455 A". [read: p.11; the "22" is an OCR/typesetting garble: 2 pi a0 / alpha with alpha = 1/137 gives 455.6 A; 4 a0 x 137 would give 290 A [calc]]. It says a cavity that suppresses ZPF at this wavelength and shorter makes the orbit decay, releasing "on the order of 1 to 10 eV" for d about 250 A, and for noble gases (He to Xe) 0.1 micron "reasonable" to release 1 to 10 eV per injection. The 1 to 10 eV figure is an assumption ("we assume"); no calculation produces it. [read: p.11]

a. Is the mathematics internally consistent? Partly. The numbers it does state are right: orbital frequency alpha c/(2 pi a0) = 6.58e15 Hz [calc] (patent's "6.6x10^1?" is garbled in OCR), wavelength 455.6 A [calc]. But the argument has internal tensions: - A cavity of d = 1000 A (0.1 micron) excludes, for ideal plates, only modes with wavelength longer than 2d = 2000 A. The relevant wavelength, 455 A, is 4 times shorter. The patent knows this and appeals to a "stairway" of partial suppression at d/n and to the tail of the electron distribution out to five Bohr radii. [read: p.11] That partial suppression is a few per cent in the field near a mirror (field correction terms go as 1/(2kz), about 7% for z = 500 A, lambda = 455 A [calc, order of magnitude]) and it oscillates in sign with wavelength, so a broadband effect nearly averages out. No calculation of the net effect is offered. - The patent itself says the extension to atoms other than hydrogen "has not yet been proven theoretically". [read: p.12] - The 2012 experimental paper uses a different suppressed wavelength (it says "0.1 micron" for Xe's outer orbital and "0.05 micron" for He, inside 0.2 micron cavities) and then finds the opposite ordering; the authors say "We do not know the reason". [read: Dmitriyeva-Moddel 2012 section 2.3.9 / 3.3]

b. Consistency with established experiments. Not consistent. The established, QED-based treatment of a ground-state atom in a 100 nm cavity is the Casimir-Polder shift. I estimated it (perfect-conductor plate, retarded formula U = 3 hbar c alpha_pol / (32 pi^2 eps0 z^4), z = 50 nm, atomic polarizabilities from [memory, medium-high], two walls approximated as double): He about 1.5e-9 eV, Ar about 1.2e-8 eV, Xe about 3e-8 eV. [calc] A non-retarded dipole-image estimate for any one-electron atom gives about 3e-9 eV. [calc] So the established shift is 8 to 10 orders of magnitude below the claimed 1-10 eV. Other cross-checks: the entire Casimir energy of the vacuum in a volume of one atom (a0^3) at d = 100 nm is only 4.3 J/m^3 x 1.5e-31 m^3 = 4e-12 eV [calc]; thermal energy kT at 300 K is 0.026 eV [calc], so 1-10 eV is 40 to 400 kT, a chemical-bond-sized change. Spectroscopy of atoms near surfaces and in cavities (Hinds, Haroche groups; Sukenik 1993) sees shifts of kHz to MHz (10^-11 to 10^-8 eV) at these distances and agrees with QED. [web: https://digitalcommons.odu.edu/physics_fac_pubs/7/; scale of shifts [memory, medium]] A 1-10 eV shift would be a gross change in the atomic spectrum of every noble gas in any 100 nm gap (for example inside zeolites, aerogels and porous glass, where noble gases adsorb routinely), and none has been seen. Also, 10 eV is comparable to ionization energies (Xe 12.1 eV, [memory, high]), which would make atoms in the cavity near-ionized.

c. Tested? Yes, in several attempts; none supports it. - Dmitriyeva and Moddel, Physics Procedia 38 (2012): He, Ar, N2, Xe through polycarbonate membranes with 0.1-0.4 micron pores, pyroelectric detector. An infrared (IR) signal was "clearly observed", but it was about 10^-7 W at the detector versus "milliwatts" predicted at 1 eV per atom, "thousands times larger than the measured power". The gas ordering was opposite to prediction (He highest, Xe lowest) and bare polycarbonate gave more than gold-coated. Their own power balance says "the power out was less than the power in and there is no need to invoke an additional power source". Power in ~2e-5 W (5 sccm, 4 torr). Their conclusion: ZPE "remains a possible explanation". [read: Dmitriyeva-Moddel 2012, section 3.1 and 4, in research/files/] Gas atoms in that experiment ran at several torr with a mean free path about 10 micron, longer than the cavity, which the authors note [read: section 3.3]. - Moddel and Dmitriyeva, Atoms 7, 51 (2019): reuse these data, call them "tantalizing but ... inconclusive", and say the lower-than-expected power is "some if not all" attributable to pore variation. [read: Atoms 2019 lines on "Deviations" and Future Work, research/files/] - Henriques MSc (Coimbra 2014): optically pre-excited atoms entering Al-coated nanopores; "some evidence" for Xe, nothing above noise with pre-excited Xe; detector judged insensitive. [web: https://estudogeral.uc.pt/handle/10316/28082 via Investigator; thesis PDF not read by me] - Puthoff, Little and Ibison (2002): a synchrotron test of Casimir-cavity ground-state shifts, "so far without success". [web: arXiv:1012.5264 via Investigator] - Haisch in 2009: "not yet been able to prove anything in the laboratory"; sporadic signals "can't be ruled out as experimental error". [web: https://www.nextbigfuture.com/2009/02/jovion-corporation-gets-patent-for-zero.html via Investigator] No independent replication of an energy release, and no positive result was replicated. Negative or null results exist from the inventors' own group and from a thesis; they are not hostile critics.

d. What would have to change if true? QED's calculation of the ground-state energy of atoms in cavities (Casimir-Polder / Lamb shift), tested at the percent level, would be wrong by ten orders of magnitude; the atomic ground state would not be the lowest-energy state; and atomic spectra in nanoporous materials would be anomalous.

Verdict, claim 1. Strong form (1-10 eV per transit): Refuted by calculation (QED shift about 1e-9 to 3e-8 eV). Observation: Tested, not supported. Weak form (some IR emission from gas flowing through membranes): observed, but ordinary explanations (Joule-Thomson, friction, heat conduction, emissivity of the membrane) were not excluded by the authors; this part is Never properly tested.

Claim 2. Re-absorption on exit; continuous net power, "extracted locally, replenished globally"

Argument [read: p.12; Atoms 2019 section 2.3]. On leaving the cavity the electron "would absorb energy from the zero-point field and be re-excited", so a flowing gas cycles and the absorber gets the energy each time.

a. Mathematical consistency. No equations; the cycle is described verbally. A physical consistency problem exists inside the argument. Whatever sets an atom's energy as a function of position is a potential energy U(z). If atoms entering lose energy |U| (to light or heat), then climbing out requires |U| to be supplied; in a static geometry the energy gained on entry is paid back on exit. The authors themselves name this: "one might argue ... they would have to surmount a potential hill to exit the cavity ... Still, we cannot totally rule out this potential hill argument." [read: Atoms 2019, section 2.3.8] They also say that if the entrance is slower than the equilibration time, "the entire process occurs in quasi-equilibrium and detailed balance applies ... there would be no excess energy to shed", and conclude "it is not clear whether equilibrium considerations render it incapable of providing harvestable energy." [read: Atoms 2019 section 2.3.10]. Detailed balance (in equilibrium each process is matched by its reverse at equal rate) is the standard reason such closed cycles at one temperature cannot give net work. Here everything is at room temperature, the ZPF is the same inside and outside except at the walls, and a gas at 300 K has thermal speeds of about 1260 m/s (He) and 220 m/s (Xe) [calc], crossing a 10 micron stripe in 8 to 45 ns while the drift is 0.1 m/s [calc]. The atoms are thermalized with the walls on the time scale of nanoseconds, many times per second per atom, far faster than the flow. That is the equilibrium situation the authors worry about.

b. Consistent with experiments? The Casimir-Polder force is a conservative force and no cyclic energy gain from a static cavity has been seen in cavity-QED, atom-optics or MEMS work [memory, medium]. Casimir's own result is conservative, and attempts to extract net work from static Casimir geometries have always failed on the return stroke. Forward himself called the field "probably conservative" [read: Forward1984 p.1-3]. The "replenished globally" language does not change the logic: the vacuum field has no spare energy to donate, because its ground state is the lowest-energy state of the field; energy cannot be removed from the lowest state without a lower state to move to. [memory, high; standard QFT]

c. Tested? See claim 1; no net-power experiment was run. The 2012 experiment measured only radiation, not whether the absorber's energy output exceeded the pump power; the authors found the opposite (out < in). Power-out never exceeded power-in in any report I have seen. Funding and follow-up: DARPA (via SPAWAR grant N66001-06-1-2026), University of Colorado proof-of-concept money and private money (about $200k by 2009; about $10M sought) [web via Investigator]. After 2012 no further gas-flow experiment by the group was found; Jovion's 2023 website has no progress posted. [web via Investigator]

Absence of evidence (record goes quiet), competing explanations. Investigator lists: (a) the effect did not hold up and the work was dropped quietly; (b) funding ran out; (c) trade secrecy; (d) the idea moved into Moddel's MIM (metal-insulator-metal) Casimir-cavity devices from 2019. My weighing: the physics argument above is itself enough to predict (a) and (b) without any suppression; (d) is documented (Moddel et al., Phys. Rev. Research 3, L022007, 2021; patent family US 11,133,758). No evidence for (c) beyond silence; the 2012 and 2019 papers are public and candid about the weak results, which argues against concealment of positive data. Records that would decide: USPTO maintenance-fee history (a lapse flagged "2011-12" on Google Patents is unresolved), DARPA/DTIC final report for the SPAWAR grant, Hathaway Research International (HRI) project report, Jovion filings.

d. What would change. The second law/detailed balance for a system at a single temperature; and the ground-state character of the vacuum.

Verdict, claim 2. Refuted by calculation / equilibrium argument (conservative potential plus detailed balance), in the strong form of continuous net power from room-temperature gas cycling. Weak form (non-equilibrium flow might evade detailed balance): Never properly tested; the authors flag it as an open question and the only data on it show power out less than power in.

Claim 3. SED explains atomic stability; the hydrogen balance extends to all atoms

a. Mathematics. The classical Larmor power at the Bohr orbit is 4.67e-8 W (2.9e11 eV/s), so without replenishment the electron loses 13.6 eV in 4.7e-11 s [calc], matching the patent's "less than one-trillionth of a second". The harmonic-oscillator version (Boyer 1975, Puthoff 1987) gives a Bohr-sized balance, which Haisch himself calls "a weakness" because a Coulomb orbit is not a harmonic oscillator, and he lists an autoionization cutoff problem. [web: http://www.calphysics.org/zpe.html via Investigator] Cole and Zou's simulations (2003-2004) did produce a ground-state-like distribution for a certain time. [read: patent p.11; Cole and Zou not read by me]

b. Established results. Later work from other SED groups finds hydrogen self-ionizes in SED: Nieuwenhuizen and Liska (2015) confirm "that the hydrogen self-ionizes in Stochastic Electrodynamics" and that relativistic corrections leave it present; they propose an added 1/r^2 term as a remedy. [web: https://philpapers.org/rec/NIESOT-2; https://arxiv.org/pdf/1506.06787] Moddel and Dmitriyeva also cite Cetto and de la Peña (1988) as showing that "under particular simplifying assumptions this [cavity] shift is not predicted" [read: Atoms 2019, text around the Cetto citation (ref 46)]: a point against claim 1 from within SED, which I did not read in the original. SED generally has not reproduced atomic structure for multi-electron atoms, quantum interference, or entanglement (Bell tests); standard quantum mechanics (QM) does all of these. [memory, high on the general statement, medium on SED-specific details]

c. Tested. SED hydrogen only by simulation; no experiment distinguishes SED from QM for atoms. Haisch noted a weakness. The patent's own text says extension to other atoms is "not yet proven theoretically". [read: p.12]

d. If true. SED would replace QM as the explanation of atomic structure. It would have to reproduce the Pauli exclusion pattern, the periodic table, selection rules and photon statistics; none are derived.

Verdict, claim 3. Historically superseded but instructive (a worthwhile classical attempt, now known to have trouble even for hydrogen, with the multi-electron extension used by the patent being an unproven analogy). The extension step "all atoms" is Not testable as stated in the patent as written (no theory to make a prediction). Learner's note: the idea that vacuum fluctuations sustain atoms is not entirely crazy; in QED the vacuum field does matter (spontaneous emission, Lamb shift), but it does not hold the ground state up, since the ground state cannot radiate.

Claim 4. The patent's power estimates

I reproduced the arithmetic from the stated assumptions [calc] (details checked with the deep read of Patent7379286, which I verified against my own numbers): - Channels: 10x10 cm plates, 0.1 micron gap, gas at 10 cm/s at 1 atm: 1e-8 m^2 x 0.1 m/s x 2.69e25 /m^3 x 5000 strips = 1.34e20 transitions per second; x 1 eV = 21.5 W, x 10 eV = 215 W. Matches "21 to 210 W". The result is simply atoms per second x the assumed 1-10 eV. [read: patent p.13] - Tunnels (1 cm^3 cube): 2.7e19 atoms per second x 500 transitions each = 1.3e22 per second; x 1 eV = 2160 W. Matches "2150 to 21500 W". [read: p.14] - Bonded-wafer device: "1 to 10 watts at 8 atmospheres" [read: p.15]; no calculation of that figure appears in the patent. - Pumping: the patent scales Roy's data to "a pressure of 1700 Pa, corresponding to 17 atmospheres" and "1.7 milliwatts". 1700 Pa is 0.017 atm; 17 atm is about 1.7e6 Pa, and 1.7e6 Pa x 1e-6 m^3/s = 1.7 W, not 1.7 mW [calc]. The figures differ by a factor of about 1000. Even the 1.7 W figure would be far below the claimed 2.2-22 kW, so the conclusion "more power is produced than used" does not rely on the typo, but the comparison made is against the large tunnel estimate, not the bonded-wafer estimate (1-10 W at 8 atm), for which 1.7 W would not leave a net gain. The pressure drop is also scaled linearly in thickness although the flow at 200 nm pores is in the free-molecular regime (Knudsen number: mean free path 40-190 nm at 1 atm [calc] versus 100-200 nm gaps), where the flow-versus-pressure relation is different. Roy's citation is also given as 1993 in the body text but 2003 in the reference list. [read: patent p.2 and p.15] - Water bath: 1 mm of water with absorption coefficient 10 cm^-1 absorbs 1 - e^-1 = 63%, not "nearly all". [calc; read: p.14-15]

a-d. (a) Arithmetic is consistent with the premise, with the pumping figure internally inconsistent. (b) The premise (1-10 eV) is refuted under claim 1. (c) No power output measured. The 2012 experiment measured power at the detector of ~1e-7 W against a predicted milliwatts, and power in exceeded power out. (d) Same as claim 1.

Verdict, claim 4. Internally inconsistent (pumping figures); the output estimates are consistent with their assumption but rest on a Refuted-by-calculation premise.

Claim 5. Forward (1984) vacuum-fluctuation battery

Forward, Phys. Rev. B 30, 1700 (1984): a stack of charged conducting foils, held apart electrostatically just short of the Casimir force, allowed to collapse, doing work; recharge by pushing apart. He treats the field as probably conservative and an analogy with a hydroelectric dam; he gives no power estimate. [read: Forward1984 deep read]

a. Mathematics. I checked the coefficients: pi h c / 1440 per unit area energy and pi h c / 480 pressure with h Planck's constant (not hbar) equal pi^2 hbar c/720 and pi^2 hbar c/240. [calc via SymPy in deep read; I rechecked: hbar = h/2pi gives these identically] Sign of Eq. 2 is dropped in print (a typo); magnitude right. "980" in OCR is 480 on the page image. Dilute Lifshitz factor 0.35 is right in the limit [calc in deep read].

b-c. The Casimir force itself is Established (Lamoreaux 1997; Mohideen and Roy 1998 [memory, high]). The battery is a storage device: the energy released on collapse is less than or equal to what is spent to pull the foils apart again; Forward calls it a battery, not a source. It was never built [no source found for a build]; at 1-20 nm separations, material properties, roughness and stiction dominate, which he notes. [read: Forward1984]

d. Nothing in mainstream physics needs to change.

Verdict, claim 5. The Casimir force part: Established. The device as an energy store: Plausible, untested. The patent's use of it ("a simple, but not practical, energy extraction experiment") is a misreading: Forward's scheme returns what was put in.

3. Investigator's findings, weighed

• 2012 and 2019 experimental papers (inventors' own tests): kept; they are the most important data: IR signal real, ordering opposite to prediction, magnitude about 1e-3 or lower of prediction, power out less than power in; authors list conventional explanations that "do not clearly fit" but "a combination of effects could". I judge ordinary heating (Joule-Thomson/viscous heating in the membrane, outgassing and adsorption, heat conduction in the holder) most plausible; no control with a non-Casimir pore size (above about 1 micron) or with a Casimir-free membrane geometry is reported beyond the 3 mm hole test. Neutral-test status: a controlled test was attempted by interested parties; it did not support the scheme.
• Haisch's own caveats (calphysics.org, 2009 interview): kept; strong honest self-assessment.
• Henriques 2014 (Coimbra) and Puthoff-Little-Ibison 2002: independent attempts, null or inconclusive; kept. I did not read their PDFs or the thesis; I rely on the Investigator and the 2019 paper's summary.
• Cetto and de la Peña 1988: SED analysis finding the shift not predicted under simplifying assumptions; noted from the 2019 paper's citation; original not read.
• Moddel's MIM line (2019-2021): different mechanism; projected "up to 250 W" for a 10x10 cm array; estimates, not measurements. Moddel's own 2019 review says rectifier-type and mechanical schemes violate thermodynamic constraints. [read: Atoms 2019; web via Investigator] I did not review the MIM claims here.
• Hydrino/BlackLight link, third-party patents citing Haisch-Moddel: related only by citation or commenter's opinion; no evidence bearing on physics here.
• Kotov et al. 2025 (Casimir-Lifshitz cavity modification of ground states): unread; its position on the patent is unknown. Its existence shows mainstream theory does treat cavity-modified ground states, at the CP scale.
• Patent status, maintenance-fee lapse, funding chain, Jovion/HRI: unresolved; see claim 2's absence-of-evidence analysis.

4. Corrections to Step 1

1. claims.md claim 1 states the shift as the Patent's own "Casimir-Lamb shift"; fine, but the patent derives 1-10 eV by assumption ("we assume"), which should be stated. Not a Step 1 error, a precision point.
2. facts and deep-read describe "220.529137 = 455 A" as is; the product 220.529137 is actually 290 A. The patent's 455 A requires 2 pi in place of 2*2. Treat the printed expression as an OCR or typesetting garble; the number 455 A is correct for 2 pi a0/alpha. [calc]
3. Deep read says the output is "1700 Pa" vs "17 atmospheres" inconsistent: confirmed [calc]. I add that the 17-atm (1.7 W) reading would still not give net power for the 1-10 W bonded-wafer device the patent calls its preferred embodiment, because the comparison made is with the larger cube estimate.
4. The deep read's note that the 2012 paper's own power balance concluded out < in is not highlighted in claims.md; it should be: "At this point it appears that the power out was less than the power in and there is no need to invoke an additional power source" [read: 2012, section 3.1]. The Investigator's lead gives this accurately ("Measured power in exceeded the detected power out").
5. Investigator: the Investigator's date for the SSE abstract is an inference; I did not check.
6. facts and claims correctly flag that the patent and abstract come from interested parties.

5. What would settle it

• Spectroscopy of Xe or Ar in a precisely defined 100 nm slit (not a random polycarbonate membrane): a 1-10 eV shift would be glaring; the QED shift is about 1e-8 eV.
• A calorimetric measurement of heat out versus pump power with a matched control (large-pore membrane, same gas, same flow) and with a thermal reference: net power out > power in.
• A theoretical derivation, not an assumption, of the energy released by an atom entering a Casimir slit, from either SED or QED, with its exit cost included.

6. Web checks done (budget about 6; used 2)

• Nieuwenhuizen and Liska: SED hydrogen self-ionization, retagged [web] https://philpapers.org/rec/NIESOT-2 and https://arxiv.org/pdf/1506.06787 (search snippets read; full text not read).
• Sukenik et al. 1993: [web] https://digitalcommons.odu.edu/physics_fac_pubs/7/ (abstract-level).
• Not web-checked (remain [memory]): atomic polarizabilities (medium-high; they only matter to an order of magnitude and a factor of 10 error would not change the conclusion), the size of CP shifts in atom-cavity spectroscopy (medium), Lamoreaux and Mohideen-Roy Casimir measurements (high). The calculation of CP shift is mine and uses the textbook retarded formula, [memory, high] for the formula.